Batch vector embeddings

Opensolr API Endpoint: `batch_embed`

Overview

The batch_embed endpoint allows you to generate vector embeddings for multiple text payloads in a single request — up to 50 items at a time, each up to 50,000 characters. This is significantly faster than calling the single embed endpoint in a loop, because all texts are encoded in a single GPU/CPU batch.

Endpoint URL

https://api.opensolr.com/solr_manager/api/batch_embed

Supports GET and POST requests.

Authentication Parameters

Parameter	Type	Required	Description
email	string	Yes	Your Opensolr registration email address.
api_key	string	Yes	Your API key from the Opensolr dashboard.
index_name	string	Yes	Name of your Opensolr index (for authentication).

Embedding Parameters

Parameter	Type	Required	Description
payloads	JSON array	Yes	A JSON array of strings to embed. Max 50 items, each max 50,000 characters.

You can pass payloads as:

A GET/POST parameter: payloads=["text1","text2","text3"]
A JSON body: {"payloads": ["text1", "text2", "text3"]}

Rate Limits

Limit	Value
Max items per batch	50
Max characters per item	50,000
Min characters per item	2

Query vs Document Embedding\n\nSet `is_query=1` when embedding search queries (user input at search time). Leave `is_query=0` (default) when embedding documents/passages at index time. The model prepends an instruction prefix for queries that optimizes the vectors for retrieval — improving recall by 5–15%, especially on natural language queries and cross-language matching.\n\n---\n\n## Example Response

{
  "embeddings": [
    [0.0123, -0.0456, ...],
    [0.0789, -0.0012, ...],
    [0.0345, -0.0678, ...]
  ],
  "dimension": 1024,
  "count": 3
}

Quick Test (1-liner)

curl -s "https://api.opensolr.com/solr_manager/api/batch_embed?email=&api_key=LOGIN_FOR_API_KEY&index_name=your_index&payloads=%5B%22hello+world%22,%22opensolr+is+great%22%5D&is_query=0"

PHP Example

<?php
$url = "https://api.opensolr.com/solr_manager/api/batch_embed";

$payloads = [
    "Distributed search across multiple shards",
    "Real-time indexing with soft commits",
    "Faceted navigation for e-commerce catalogs"
];

$params = [
    "email"      => "your@email.com",
    "api_key"    => "YOUR_API_KEY",
    "index_name" => "your_index",
    "payloads"   => json_encode($payloads),\n    "is_query"   => "0"  // Set to "1" for search queries
];

$ch = curl_init($url);
curl_setopt_array($ch, [
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_POST           => true,
    CURLOPT_POSTFIELDS     => http_build_query($params),
    CURLOPT_TIMEOUT        => 120
]);

$response = json_decode(curl_exec($ch), true);
curl_close($ch);

if (isset($response["embeddings"])) {
    echo "Got " . $response["count"] . " embeddings of dimension " . $response["dimension"] . "
";
    foreach ($response["embeddings"] as $i => $vector) {
        echo "Embedding $i: [" . implode(", ", array_slice($vector, 0, 5)) . ", ...]
";
    }
} else {
    echo "Error: " . ($response["error"] ?? "Unknown") . "
";
}

Python Example

import requests
import json

url = "https://api.opensolr.com/solr_manager/api/batch_embed"

payloads = [
    "Distributed search across multiple shards",
    "Real-time indexing with soft commits",
    "Faceted navigation for e-commerce catalogs",
]

params = {
    "email": "your@email.com",
    "api_key": "YOUR_API_KEY",
    "index_name": "your_index",
    "payloads": json.dumps(payloads),\n    "is_query": "0",  # Set to "1" for search queries
}

response = requests.post(url, data=params, timeout=120).json()

if "embeddings" in response:
    print(f'Got {response["count"]} embeddings of dimension {response["dimension"]}')
    for i, vec in enumerate(response["embeddings"]):
        print(f"Embedding {i}: {vec[:5]}...")
else:
    print(f'Error: {response.get("error", "Unknown")}')

JSON Body Example (alternative)

You can also send the payloads as a JSON body:

curl -s -X POST "https://api.opensolr.com/solr_manager/api/batch_embed" \
  -H "Content-Type: application/json" \
  -d '{
    "email": "your@email.com",
    "api_key": "YOUR_API_KEY",
    "index_name": "your_index",
    "payloads": ["hello world", "opensolr is great", "batch embeddings rock"],\n    "is_query": "0"
  }'

Solr Schema Requirement

To store batch embeddings in Solr, your schema needs a vector field:

<field name="embeddings" type="vector" indexed="true" stored="false" multiValued="false"/>
<fieldType name="vector" class="solr.DenseVectorField" vectorDimension="1024" required="false" similarityFunction="cosine"/>

⚠️ This API uses the intfloat/multilingual-e5-large-instruct embedding model which produces 1024-dimensional vectors.

Use Cases

Bulk preprocessing: Embed hundreds of documents before indexing them into Solr.
Comparison: Generate embeddings for multiple candidates and compare them using cosine similarity.
Clustering: Embed a set of texts and feed the vectors into a clustering algorithm.
Caching: Pre-compute embeddings for frequently searched queries.

Support

For more information or help, visit Opensolr Support or use your Opensolr dashboard.

Custom Plans Available

This is a premium feature available on custom plans tailored to your needs and budget. For small websites, we can even provide these features for free after validating your use case. Contact us at support@opensolr.com to discuss your requirements.